Case Study: Traffic Forecasting in Bank Branches

Context

• Canadian financial institution
• Four branches in a medium-sized town

Problem

• Forecast traffic in each of four branches in order to determine the adequate number of tellers at each time of the day
• High traffic variability depending on the time of the day, the day of the week, and whether government social support cheques are being sent out
• Historical data apply to seven small branches that are being merged into four larger branches; forecasts must be made for the new branches

Solution

• ApSTAT developed a forecasting model accounting for:
  • The expected traffic in each branch, in 15-minute intervals, varying according to daily, weekly and monthly seasonalities
  • The average time to process transactions of different types, which allows to determine the correct number of tellers from the expected traffic
• The model can adapt to new branches for which no historical data has been collected
• The model can incorporate a quality-of-service objective, for instance to ensure that clients do not wait more than 3 minutes, 80% of the time

Benefits

• Better customer service by having the right number of tellers, at the right time, in the right branches
• Maximization of technological investments: increase profitability by exploiting accumulated historical transaction and branch traffic data

« Back to Case Studies